DOCSLIB.ORG

(12) United States Patent (10) Patent No.: US 8,768,313 B2 Rodriguez (45) Date of Patent: Jul

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
(12) United States Patent (10) Patent No.: US 8,768,313 B2 Rodriguez (45) Date of Patent: Jul USOO8768313B2 (12) United States Patent (10) Patent No.: US 8,768,313 B2 Rodriguez (45) Date of Patent: Jul. 1, 2014 (54) METHODS AND SYSTEMS FOR IMAGE OR (56) References Cited AUDIO RECOGNITION PROCESSING U.S. PATENT DOCUMENTS (75) Inventor: Tony F. Rodriguez, Portland, OR (US) 7,272.407 B2 * 9/2007 Strittmatter et al. .......... 455,500 7.925,265 B2 * 4/2011 Souissi ......................... 455,445 (73) Assignee: Digimarc Corporation, Beaverton, OR 8,229, 160 B2 7/2012 Rosenblatt (US) 8,358,811 B2 * 1/2013 Adams et al. ................. 382,118 2002/0102966 A1 8, 2002 Lev et al. (*) Notice: Subject to any disclaimer, the term of this 2004/0212630 A1 10/2004 Hobgood et al. patent is extended or adjusted under 35 2006/00316842005/0091604 A1 4/20052/2006 SharmaDavis et al. U.S.C. 154(b) by 819 days. (Continued) (21) Appl. No.: 12/855,996 OTHER PUBLICATIONS (22) Filed: Aug. 13, 2010 Ahmed, et al. MACE—Adaptive Component Management Middleware for Ubiquitous Systems, Proc. 4th Int'l Workshop on (65) Prior Publication Data Middleware for Pervasive and Ad-Hoc Computing, 2006. US 2011 FO143811 A1 Jun. 16, 2011 (Continued) Primary Examiner — Blane JJackson Related U.S. Application Data (74) Attorney, Agent, or Firm — Digimarc Corporation (60) Provisional application No. 61/234,542, filed on Aug. 17, 2009. (57) ABSTRACT Many of the detailed technologies are useful in enabling a (51) Int. Cl. Smartphone to respond to a user's environment, e.g., so it can H04M 3/42 (2006.01) serve as an intuitive hearing and seeing device. A few of the G06K 9/00 (2006.01) detailed arrangements involve optimizing division of shared H04N I/00 (2006.01) processing tasks between the phone and remote devices; H04L 29/08 (2006.01) using a phone GPU for exhaustive speculative execution and G06K 9/22 (2006.01) machine vision purposes (including facial recognition); novel (52) U.S. Cl. device architectures involving abstraction layers that facili CPC ....... H04N I/00127 (2013.01); H04N I/00244 tate substitution of different local and remote services; inter (2013.01); H04N I/00331 (2013.01); H04N actions with private networks as they relate to audio/image 1/00328 (2013.01); H04L 67/16 (2013.01); processing; adapting the orders in which operations are G06K 9/228 (2013.01); G06K 9/00986 executed, and the types of data that are exchanged with (2013.01); H04N2201/0084 (2013.01); G06K remote servers, in accordance with current context; reconfig 9/00288 (2013.01); H04N I/00336 (2013.01) uring networks based on sensed social affiliations among USPC ......... 455/4.14.1; 455/415; 382/276; 382/118 users and in accordance with predictive models of user behav (58) Field of Classification Search ior, etc. A great variety of other features and arrangements are USPC ......... 455/4.1.2, 414.1, 66.1, 556.1, 557, 415; also detailed. 382/118, 173, 190, 276 See application file for complete search history. 16 Claims, 55 Drawing Sheets USERSNAPS IMAGE FACIAL oct FFT WAWELT RCCGNTION TRANSFORTRANSFORMITRANSFORM t EDGE EIGENWALUE SPECTRUM OCR ETECTON ANALYSIS ANALYSIS g es Atter FOURIER-MELLIN TEXTURE COLOR 2. XTRACTION TRANSFORM CLASSFR ANALYSS g GIST EMOTION BARCODE AGE se g PROCESSNS CLASSIFIER DECODER DETECTION s WATERMARK ORIENTATION I SIMILARITY GENDER ECOER ETECTION PROCESSORETECTON 3 OBJECT METADATA GEOLOCATION | ETC 3. SEGMENTATON ANAYSIS PROCESSENG Cs ROUTER SRWC SRWC SERVICE Row}ER PRWR2 PROWIRN CELLPHONE USERNTERFACE US 8,768,313 B2 Page 2 (56) References Cited neering for Self-Adaptive Systems, Springer Berlin Heidelberg, Jun. 19, 2009, pp. 146-163. U.S. PATENT DOCUMENTS Gu, Adaptive offloading inference for delivering applications in per 2006,0047584 A1 3, 2006 Vaschillo et al. vasive computing environments, Proc IEEE Int’l Confon Pervasive 2006/0217199 A1 9, 2006 Adcox et al. Computing and Communication, 2003. 2007/010.0480 A1 5, 2007 Sinclair et al. Kirsch-Pinheiro, , et al. Context-Aware Service Selection Using 2007/O159522 A1 7, 2007 Neven Graph Matching, 2nd Non Functional Properties and Service Level 2009 OO31381 A1 1/2009 Cohen et al. Agreements in Service Oriented Computing Workshop (NFPSLA 2009/0237546 A1 9, 2009 Bloebaum et al. SOCO8), CEUR Workshop Proceedings, vol. 411, 2008. 2009,0252383 A1 10/2009 Adam et al. 2009,0279794 A1 11/2009 Brucher et al. Noble, etal, Agile Application-Aware Adaptation for Mobility, Proc. 2009/0285492 A1 1 1/2009 Ramanujapuram et al. of the ACM Symposium on Operating System Principles (SOSP), 2010/0260426 A1 10/2010 Huang et al. 1997. Osman, et al. The Design and Implementation of Zap: A System for OTHER PUBLICATIONS Migrating Computing Environments, Proc of the Fifth Symposium Balan, Tactics-Based Remote Execution for Mobile Computing, on Operating Systems Design and Implementation (OSDI), 2002. MobiSys 2003. Reichle, et al. A Comprehensive Context Modeling Framework for Chun, Augmented Smartphone applications through clone cloud Pervasive Computing Systems, Distributed Applications and execution, Proc. of the 8th Workshop on Hot Topics in Operating Interoperable Systems, Springer Berlin Heidelberg, 2008. Systems, May 2009. Zachariadis, et al. Adaptable Mobile Applications: Exploiting Logi Flinn et al. Balancing Performance, Energy, and Quality in Pervasive cal Mobility in Mobile Computing, in Mobile Agents for Telecom Computing, Proc. of the 22nd International Conference on Distrib munication Applications, Springer Berlin Heidelberg, 2003, pp. 170 uted Computing Systems (ICDCS), Jul. 2002. 179. Flinn, et al. Self-Tuned Remote Execution for Pervasive Computing, Zachariadis, et al. Building Adaptable Mobile Middleware Services Proc. of the 8th Workshop on Hot Topics in Operating Systems Using Logical Mobility Techniques, in Contributions to Ubiquitous (HotOS), May 2001. Computing, Springer Berlin Heidelberg, 2007, pp. 3-26. Geihs, et al. Modeling of Context-Aware Self-Adaptive Applications in Ubiquitous and Service-Oriented Environments, Software Engi * cited by examiner U.S. Patent US 8,768,313 B2 U.S. Patent Jul. 1, 2014 Sheet 2 of 55 US 8,768,313 B2 Z"SOIH LOETEO U.S. Patent US 8,768,313 B2 XIÈHOWALEN SSETTE?-HIAA U.S. Patent Jul. 1, 2014 Sheet 4 of 55 US 8,768,313 B2 KEYVECTOR DATA (PIXELS & DERIVATIVES) FIG. 4 RESULTA RESULT B RESULT C KEYVECTOR FG. 4A ADDRESSING, ETC / PXEL PACKET U.S. Patent Jul. 1, 2014 Sheet 5 of 55 US 8,768,313 B2 F.G. 4B TASKA TASK B TASK C TASK D WSUA TASK COMMONALY PIE CHARTS I FFT RE-SAMPLNG PDF417 BARCOOE READING F.G. 5 U.S. Patent Jul. 1, 2014 Sheet 6 of 55 US 8,768,313 B2 RESDEN CAL-UP VSUAL PROCESSING SERVICES ON ON OFF OFF ON OFF COMMON SERVICES SORTER OW LEVE PXEL PRO- 8. CESSENG LIBRARY FOW GATE CONFIGURATION, SOFTWARE PROGRAMMING EXERNAL PXEL KEYVECTOR DATA SENSOR milier HARDWARE (PROCD PIXELS & ROUTING OTHER) F.G. 6 NERNA U.S. Patent US 8,768,313 B2 U.S. Patent US 8,768,313 B2 U.S. Patent Jul. 1, 2014 Sheet 9 Of 55 US 8,768,313 B2 SOAC U.S. Patent Jul. 1, 2014 Sheet 10 Of 55 US 8,768,313 B2 >JOSSEOORHd ?JOSSEOORHd ENOHdTTHO 9NISSEOO}}d U.S. Patent US 8,768,313 B2 U.S. Patent Jul. 1, 2014 Sheet 12 Of 55 US 8,768,313 B2 MY PROFILE, INCLUDING WHOI AM, AND WHAT DO GENERALLY LIKE/WANT MY CONTEXT, INCLUDING WHERE AM, AND WHAT ARE MY CURRENT DESRES MY WISUAL OUERYPXELS THEMSELVES PACKETS RESULTS VISUAL QUERY PACKET PACKETS RESULTS PACKETS RESULTS CONFIGURABLE ARRAY OF OFFEREDATAUCTION TO AN SERVICE PROVIDERS ARRAY OF SERVICE PROVIDERS FIG. 11 U.S. Patent US 8,768,313 B2 EOLAECI>JESÍTI U.S. Patent Jul. 1, 2014 Sheet 14 of 55 US 8,768,313 B2 12 SENSOR 14 PROCESSING FOR OBJECT DENTIFICATION PROCESSING FOR HUMAN WISUAL ADDRESSING SYSTEM (e.g., JPEG compression) 15 FIG. 13 16 56 CAMERA STAGE 1 STAGE 2 STAGE Y INSTRUCTIONS INSTRUCTIONS INSTRUCTIONS ' ' ' INSTRUCTIONS 55 58a 58b. STAGE Z NSTRUCTIONS DATA 56 59 FIG. 17 U.S. Patent Jul. 1, 2014 Sheet 15 Of 55 US 8,768,313 B2 CEL PHONE FIG. 14 APPLICATION NCELL PROCESSING C) PHONE EXTERNAL FROM CELL PHONE APPLICATION l PHONENCELL APPLICATION - - - - - - - - - - - - - - - - - - - - - - a EXTERNAL FROM 10 CELL PHONE APPLICATION EXTERNAL FROM CELL PHONE U.S. Patent Jul. 1, 2014 Sheet 16 of 55 US 8,768,313 B2 0 U.S. Patent Jul. 1, 2014 Sheet 17 Of 55 US 8,768,313 B2 L CD a. O -NO(HONAS NO||W/Z1 U.S. Patent Jul. 1, 2014 Sheet 18 Of 55 US 8,768,313 B2 8/ 01 --------------------------;;;;;;;. U.S. Patent Jul. 1, 2014 Sheet 20 Of 55 US 8,768,313 B2 NOISIATWO ?JEHLO SEKOLARJES CIEX1 ?JEAN-JES -----Z___---------Z___---- TOOOLO>jc) U.S. Patent Jul. 1, 2014 Sheet 21 Of 55 US 8,768,313 B2 RESPONSE TIME ROUTING NEEDED CONSTRANTS OPERATION DETALS USER HARDWARE • High level PREFERENCES o What special purpose operations hardware is local? O LOW level o What is Current operations hardware utilization? o Common operations o CPU pipeline length o Operations that are preconditions to o Pipeline stall risks others CONNECTWTY o Power consumption STATUS GEOGRAPHICAL CONSIDERATIONS RULES, HEURISTICS NFORE REMOTE PROVIDER(S), E.G., Readiness Speed Cost Attributes of importance to user OPERATIONS FOR OPERATIONS FOR LOCAL PROCESSING REMOTE PROCESSENG FIG. 19B U.S. Patent Jul. 1, 2014 Sheet 22 Of 55 US 8,768,313 B2 MICRO MIRROR PROJECTOR U.S. Patent Jul. 1, 2014 US 8,768,313 B2 U.S. Patent Jul. 1, 2014 Sheet 24 Of 55 US 8,768,313 B2 U.S. Patent Jul. 1, 2014 Sheet 25 Of 55 US 8,768,313 B2 CAPTURE OR RECEIVE IMAGE MAGE AND/OR METADATA - - - - - - - - - - - - - - - - PRE-PROCESS IMAGE, SUBMIT TO GOOGLE, RECEIVE e.g., IMAGE ENHANCEMENT, AUXLIARY INFORMATION : OBJECT SEGMENTATION/ : EXTRACTION, etc.
Load more

Recommended publications
  • 0321741331.Pdf
    THE PHOTOSHOP ELEMENTS 9 BOOK FOR DIGITAL PHOTOGRAPHERS The Photoshop Elements Published by 9 Book for Digital New Riders Photographers Team Copyright ©2011 by Scott Kelby CREATIVE DIRECTOR Felix Nelson First edition: November 2010 TECHNICAL EDITORS All rights reserved. No part of this book may be reproduced or transmitted in any Cindy Snyder form, by any means, electronic or mechanical, including photocopying, recording, Kim Doty or by any information storage and retrieval system, without written permission from the publisher, except for inclusion of brief quotations in a review. TRAFFIC DIRECTOR Kim Gabriel Composed in Frutiger, Lucida, and Apple Garamond Light by Kelby Media Group, Inc. PRODUCTION MANAGER Dave Damstra Trademarks All terms mentioned in this book that are known to be trademarks or service marks ART DIRECTOR have been appropriately capitalized. New Riders cannot attest to the accuracy of Jessica Maldonado this information. Use of a term in the book should not be regarded as affecting the validity of any trademark or service mark. COVER PHOTOS COURTESY OF Photoshop Elements is a registered trademark of Adobe Systems, Inc. Scott Kelby Windows is a registered trademark of Microsoft Corporation. Matt Kloskowski Macintosh is a registered trademark of Apple Inc. iStockphoto.com Warning and Disclaimer This book is designed to provide information about Photoshop Elements for digital photographers. Every effort has been made to make this book as complete and as accurate as possible, but no warranty of fitness is implied. The information is provided on an as-is basis. The authors and New Riders shall have neither the liability nor responsibility to any person or entity with respect to any loss or damages arising from the information contained in this book or from the use of the discs or programs that may accompany it.
    [Show full text]
  • The Spirit of Innovation Why Innovation and Value in Japan? Why Now?
    THE IAFOR GIVS-Tokyo2018 | Programme & Abstract Book | Programme & GIVS-Tokyo2018 GLOBAL INNOVATION & VALUE SUMMIT 2018 PROGRAMME & ABSTRACT BOOK THE SPIRIT OF INNOVATION Why Innovation and Value in Japan? Why Now? ISSN:2433-7544 (Online)ISSN: | 2433-7587 (Print) イノベーションの精神： 日本からのインスピレーション なぜ日本でのイノベーションと価値創造なのか？なぜ今なのか？ Global Business Hub Tokyo, Japan | October 5, 2018 Toshi Center Hotel, Japan | October 6-7, 2018 Organised by IAFOR in association with the IAFOR Research Centre at Osaka University and IAFOR’s Global University Partners 開催場所・日時：グローバルビジネスハブ東京 | 2018年10月5日 開催場所・日時：都市センターホテル | 2018年10月6 〜７日 大阪大学大学院国際公共政策研究科 IAFOR 研究センター ＆IAFOR グローバル・ユニバーシティ・パートナーズ共同開催 Organised by IAFOR iafor MEDIA PARTNERS オフィシャル メディアパ ートナ ーズ CORPORATE STRATEGIC PARTNER 企業戦略パートナー SUPPORTING ORGANISATIONS 支援団体 SUMMIT GUIDE サ ミット・ガ イド Summit Theme サミットテーマ Symposium at a Glance シンポジウム・スケジュール Advisory Board 諮問委員会詳細 Organising Committee 組織委員会 Moderators モデレータ Letter of Welcome ウェルカム・レター A Few Thoughts on Innovation イノベーションとは？ Directions & Access 会場アクセス SUMMIT THEME THE SPIRIT OF INNOVATION WHY INNOVATION AND VALUE IN JAPAN? WHY NOW? 「イノベーションの精神：日本からのインスピレーション」 なぜ日本でのイノベーションと価値創造なのか？ なぜ今なのか？ Japan has been hailed as one of the world’s most creative and innovative countries, while simultaneously being maligned as a stagnant and lost economy that has lost its lustre. Yet as the world confronts the limits of Western concepts of innovation and the value that these bring, unique, sustainable and inclusive models of innovation developed throughout Japan’s long history
    [Show full text]
  • 2017 PDL Fall Update
    NEWSLETTER ON PDL ACTIVITIES AND EVENTS • FALL 2 0 1 7 http://www.pdl.cmu.edu/ PDL CONSORTIUM The PDL is 25! MEMBERS by Joan Digney, Greg Ganger & Bill Courtright Broadcom, Ltd. Citadel After a successful formative workshop in late 1992, Dr. Garth Gibson officially Dell EMC launched the PDL in 1993 with 7 students from CMU’s CS and ECE Departments. Google Having recently finished his Ph.D. research, which defined the industry standard Hewlett-Packard Labs RAID terminology for redundant disk arrays, Gibson guided PDL researchers Hitachi, Ltd. in advanced disk array research. The name “Parallel Data Lab” came from this Intel Corporation initial focus on parallelism in storage systems. In the PDL’s formative years, its Microsoft Research researchers developed technologies for improving failure recovery performance MongoDB (parity declustering) and maximizing performance in small-write intensive work- NetApp, Inc. loads (parity logging). They also developed an aggressive prefetching technology Oracle Corporation (transparent informed prefetching, or TIP) for converting serial access patterns Salesforce into highly parallel work- Samsung Information Systems America loads capable of exploit- Seagate Technology ing large disk arrays. Two Sigma The PDL first received Toshiba actual lab space at CMU Veritas (Wean Hall 3607) to go Western Digital with its name in January of 1994. As it grew, PDL CONTENTS became more spread out, with people in various ar- The PDL is 25! .............................. 1 eas of Wean Hall and the PDL News & Awards........................3 D-Level of Hamerschlag Hall. Today, PDL people Defenses & Proposals .......................4 are primarily located in Recent Publications ........................6 the Robert Mehrabian Garth and the Scotch Parallel File System Collaborative Innovation THE PDL PACKET Center (RMCIC) and the Gates Center for Computer Science.The equipment populating PDL lab spaces is often state of the art, a rare case for academic re- EDITOR search, and is upgraded frequently as technology advances.
    [Show full text]
  • AI Chips: What They Are and Why They Matter
    APRIL 2020 AI Chips: What They Are and Why They Matter An AI Chips Reference AUTHORS Saif M. Khan Alexander Mann Table of Contents Introduction and Summary 3 The Laws of Chip Innovation 7 Transistor Shrinkage: Moore’s Law 7 Efficiency and Speed Improvements 8 Increasing Transistor Density Unlocks Improved Designs for Efficiency and Speed 9 Transistor Design is Reaching Fundamental Size Limits 10 The Slowing of Moore’s Law and the Decline of General-Purpose Chips 10 The Economies of Scale of General-Purpose Chips 10 Costs are Increasing Faster than the Semiconductor Market 11 The Semiconductor Industry’s Growth Rate is Unlikely to Increase 14 Chip Improvements as Moore’s Law Slows 15 Transistor Improvements Continue, but are Slowing 16 Improved Transistor Density Enables Specialization 18 The AI Chip Zoo 19 AI Chip Types 20 AI Chip Benchmarks 22 The Value of State-of-the-Art AI Chips 23 The Efficiency of State-of-the-Art AI Chips Translates into Cost-Effectiveness 23 Compute-Intensive AI Algorithms are Bottlenecked by Chip Costs and Speed 26 U.S. and Chinese AI Chips and Implications for National Competitiveness 27 Appendix A: Basics of Semiconductors and Chips 31 Appendix B: How AI Chips Work 33 Parallel Computing 33 Low-Precision Computing 34 Memory Optimization 35 Domain-Specific Languages 36 Appendix C: AI Chip Benchmarking Studies 37 Appendix D: Chip Economics Model 39 Chip Transistor Density, Design Costs, and Energy Costs 40 Foundry, Assembly, Test and Packaging Costs 41 Acknowledgments 44 Center for Security and Emerging Technology | 2 Introduction and Summary Artificial intelligence will play an important role in national and international security in the years to come.
    [Show full text]
  • 2017 CFHT Annual Report
    2017 CFHT Annual Report Table of Contents Director’s Message……………………………………………………………………………………………... 3 Science Report ………………………………………………........................................................ 5 Hawaii Observatories Track an Interstellar Visitor…………………..…………………… 5 WIRCam and Herschel Observations of Interstellar Dust.…………………………………. 6 The Little Star That Survived a Supernova………………………....…….………………..……. 7 Rocky Planet Engulfment Explains Stellar Odd Couple......................................... 7 Capturing What Lies in the Space Between Galaxies………………………………….……. 8 Astronomers Define the Great Divide Between Stars and Brown Dwarfs………... 9 Blue Binaries Tell an Ancient Story……………………………………………………………….…. 10 Engineering Report ………..………………………………….……………………………………….……… 11 Primary Mirror Recoating – Shutdown August 2017..…………………………………..….. 11 SPIRou Development and Performance……………………………………………………………. 11 SITELLE Status………………………………………………………………………………………………….. 13 MegaCam Performance Improvements….…………………………….……………………….… 14 WIRCam Repairs….………………………….…………………………………………………………….… 17 ESPaDOnS/GRACES..………………………………………………………………….……………………. 18 Other Activity in the Engineering Group………………………….…………………………….… 20 MSE Report ……………………………………………………………………………………………………..…. 22 Summary and Overview…………………………………………………………………………………… 22 Partnership and Governance……………………………………………………………………………. 22 Science…………………………………………………………………………………………………………….. 23 Engineering………………………………………………………………………………………………………. 24 Overall Project Development and Priorities……………………………………………………… 26 Administration
    [Show full text]
  • Methods for Answer Extraction in Textual Question Answering
    \aunimo_phd" | 2007/5/25 | 3:45 | page i | #1 View metadata, citation and similar papers at core.ac.uk brought to you by CORE Department of Computer Science provided by Helsingin yliopiston digitaalinen arkisto Series of Publications A Report A-2007-3 Methods for Answer Extraction in Textual Question Answering Lili Aunimo To be presented, with the permission of the Faculty of Science of the University of Helsinki, for public criticism in Auditorium XIV, University Main Building, on June 12th, 2007, at noon. University of Helsinki Finland \aunimo_phd" | 2007/5/25 | 3:45 | page ii | #2 Contact information Postal address: Department of Computer Science P.O. Box 68 (Gustaf H¨allstr¨omin katu 2b) FI-00014 University of Helsinki Finland Email address: [email protected] (Internet) URL: http://www.cs.Helsinki.FI/ Telephone: +358 9 1911 Telefax: +358 9 191 51120 Copyright c 2007 Lili Aunimo ISSN 1238-8645 ISBN 978-952-10-3992-8 (paperback) ISBN 978-952-10-3993-5 (PDF) Computing Reviews (1998) Classification: H.3.3, H.3.4, I.2.1, I.5.4 Helsinki 2007 Helsinki University Printing House \aunimo_phd" | 2007/5/25 | 3:45 | page iii | #3 Methods for Answer Extraction in Textual Question Answering Lili Aunimo Department of Computer Science P.O. Box 68, FI-00014 University of Helsinki, Finland lili.aunimo@iki.fi PhD Thesis, Series of Publications A, Report A-2007-3 Helsinki, June 2007, 127 + 19 pages ISSN 1238-8645 ISBN 978-952-10-3992-8 (paperback) ISBN 978-952-10-3993-5 (PDF) Abstract In this thesis we present and evaluate two pattern matching based meth- ods for answer extraction in textual question answering systems.
    [Show full text]
  • Research Statement Deepak Narayanan
    Research Statement Deepak Narayanan The end of Moore’s Law has led to the rapid adoption of a number of parallel architectures, such as multicore CPUs (with SIMD), GPUs, FPGAs, and domain-specific accelerators like the TPU, each with different programming models and performance characteristics (e.g., number of cores, SIMD lane width, cache sizes). Achieving high performance on these architectures is challenging for non-expert programmers like Machine Learning engineers and data analysts, who are increasingly using computer systems to perform expensive computations (e.g., a modern ML model has trillions of floating point operations). These computations will only become more expensive going forward as ML models become larger and the amount of available data to perform analyses over increases at unprecedented rates. My research makes it easier for programmers to achieve high performance on parallel hardware for emerging work- loads such as deep neural network model training and data analytics. My main research focus has involved building systems to execute deep learning computations in a more resource-efficient and scalable way. I have examined this problem in two main contexts: a) At a microscale granularity, how should operators in a deep neural network (DNN) model be partitioned among multiple workers to maximize throughput? I built a distributed deep learning engine called PipeDream that adapts pipelining, an optimization used in conventional compilers and runtime systems, to accelerate DNN training performance with no reduction in the final accuracy of the model. PipeDream introduced pipeline-parallel training, a principled combination of pipelining with data and inter-layer model parallel training. Pipeline-parallel training is now used at Microsoft, Facebook, and Nvidia.
    [Show full text]
  • Enabling In-Memory Computation
    Processing Data Where It Makes Sense in Modern Computing Systems: Enabling In-Memory Computation Onur Mutlu [email protected] https://people.inf.ethz.ch/omutlu 10 January 2019 USC Current Research Focus Areas Research Focus: Computer architecture, HW/SW, bioinformatics, security • Memory and storage (DRAM, flash, emerging), interconnects • Heterogeneous & parallel systems, GPUs, systems for data analytics • System/architecture interaction, new execution models, new interfaces • Hardware security, energy efficiency, fault tolerance, performance • Genome sequence analysis & assembly algorithms and architectures • Biologically inspired systems & system design for bio/medicine Hybrid Main Memory Heterogeneous Persistent Memory/Storage Processors and Accelerators Broad research spanning apps, systems, logic with architecture at the center Graphics and Vision Processing Four Key Directions ◼ Fundamentally Secure/Reliable/Safe Architectures ◼ Fundamentally Energy-Efficient Architectures ❑ Memory-centric (Data-centric) Architectures ◼ Fundamentally Low-Latency Architectures ◼ Architectures for Genomics, Medicine, Health 3 A Motivating Detour: Genome Sequence Analysis 4 Our Dream ◼ An embedded device that can perform comprehensive genome analysis in real time (within a minute) ❑ Which of these DNAs does this DNA segment match with? ❑ What is the likely genetic disposition of this patient to this drug? ❑ . 5 What Is a Genome Made Of? Cell Nucleus The discovery of DNA’s double-helical structure (Watson+, 1953) 6 DNA Under Electron Microscope human
    [Show full text]
  • Beginning Microsoft Excel 2010.Pdf
    CYAN YELLOW MAGENTA BLACK PANTONE 123 C BOOKS FOR PROFESSIONALS BY PROFESSIONALS® THE EXPERT’S VOICE® IN MICROSOFT OFFICE Companion eBook Available Beginning Microsoft Excel 2010 Beginning Dear Reader, Microsoft Excel 2010 Tell me if you identify with this question: “What’s a spreadsheet?” Truth be told, this question is a very good one, and because spreadsheets are so vast, and their capabilities so broad, the answer doesn’t come easily. Abbott Katz People apply Excel spreadsheets to countless tasks, but merely knowing how to do carry out those tasks isn’t always good enough. If the user isn’t quite sure about what’s going on in the spreadsheet – and why – it makes it that much Beginning harder to carry out the tasks a second time, let alone the first – particularly if there’s no one around to ask. I’ve written Beginning Microsoft Excel 2010 with that agenda in mind. All too many introductory books slip into the click-here-click-there approach, even as they omit the answers to the “what’s going on?” questions which often lurk in the background. I’ve tried here to cover many of Excel’s capabilities but at the same time tried to anticipate the kinds of questions about them you might ask Microsoft Excel of an instructor, since you probably won’t pay for house calls. The book covers all aspects of using Excel 2010 from the day-to-day moving around a spreadsheet and performing calculations on your data, to the more complex areas of charting and using pivot tables.
    [Show full text]
  • Modeling, Embodiment, Figuration
    Jeanne Cortiel, Christine Hanke, Jan Simon Hutta, Colin Milburn (eds.) Practices of Speculation Culture & Theory | Volume 202 Jeanne Cortiel is Professor of American Studies at Bayreuth University. Her re- search interests include American science fiction, post-apocalyptic film, and glob al catastrophic risk in fiction. Christine Hanke is Chair of Digital and Audiovisual Media at Bayreuth University and conducts research in the fields of media resistance, postcolonial studies, and image theory. Jan Simon Hutta is Assistant Professor in the Cultural Geography Research Group at Bayreuth University. His research focuses on formations of power, affect and citizenship in Brazil, sexual and transgender politics, urban governmentality, and relations of subjectivity, movement and space. Colin Milburn is Gary Snyder Chair in Science and the Humanities and Professor of English, Science and Technology Studies, as well as Cinema and Digital Media at the University of California, Davis. His research focuses on the intersections of science, literature, and media technologies. Jeanne Cortiel, Christine Hanke, Jan Simon Hutta, Colin Milburn (eds.) Practices of Speculation Modeling, Embodiment, Figuration Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche National- bibliografie; detailed bibliographic data are available in the Internet at http:// dnb.d-nb.de This work is licensed under the Creative Commons Attribution-NonCommercial-NoDeri- vatives 4.0 (BY-NC-ND)
    [Show full text]
  • Live Owl Nest Box Cam
    6/24/2010 Live Owl Nest Box Cam Home Page Live Wildlife Webcams The Blog News Hunting Tips Fishing Tips Contact Us Guestbook Hunting/Fishing Videos Mallard Migration Wildlife Prints Games Download Our Toolbar Guides and Outfitters Live Owl Nest Box Cam 20320 votes Live Barn Owl Nest Box Cam - San Marcos, CA This is a live feed of a wild female barn owl named Molly. The owl box is located 15 feet off the ground. The male(McGee) has a white chest and face. He usually shows up after dark but has spent an entire day with the female. For more Information about the owls and this cam. Click Here Click Here to "Like" us on Facebook . Egg 1 2/13/10 12:15 PM Egg Broke Share 59K Egg 2 Laid 02/16/10 07:23 AM Hatched 03/21/10 11:45 AM Name: Max Egg 3 Laid 02/18/10 09:13 AM Hatched 03/23/10 6:47 PM Name: Pattison 1236 retw eet Egg 4 Laid 02/20/10 11:07 AM Hatched 03/25/10 3:11 PM Name: Austin Egg 5 Laid 02/22/10 02:00 PM Hatched 03/28/10 12:01 PM Name: Wesley Egg 6 02/25/10 09:35 AM Est. Hatch 3/28/10 Note: didn't hatch/won't hatch Find owl books and more at Amazon.com Sound and Video Help Page Barn Owl Nesting Box Attract and protect supurb hunters for rodent control the natural way www.pelletlab.com Owl Pellets & Resources No charge for shipping for next two days plus free probes & forceps.
    [Show full text]
  • The Photoshop® Elements 14 Book for Digital Photographers
    nal spine = 0.640625" The Uncover the inside tips and tricks of the trade for organizing, Scott Kelby is correcting, editing, sharpening, and retouching your photos like a pro Editor and Publisher of the highly acclaimed P Photo shop User maga- zine, and is the head of ho training for KelbyOne, t an online educational os community for photog- raphers around the world. He is co-host of ho the influential weekly photography webcast talk show The Grid, and he teaches digital p photography workshops and seminars around ® ARE YOU READY FOR AN ELEMENTS BOOK THAT BREAKS ALL THE RULES? the world. Scott is an award-winning author of Elements 14 This book breaks new ground by doing something for digital photographers that’s never more than 60 books including The Photoshop been done before—it cuts through the bull and shows you exactly “how to do it.” It’s not Book for Digital Photographers, The Lightroom a bunch of theory; it doesn’t challenge you to come up with your own settings or figure Book for Digital Photographers, and The Digital things out on your own. Instead, it does something that virtually no other Elements book Photography Book (parts 1, 2, 3, 4, and 5). has ever done—it tells you flat-out which settings to use, when to use them, and why. For more on Scott, visit his daily blog at If you’re looking for one of those “tell-me-everything-about-the-Unsharp-Mask-filter” books, http://scottkelby.com. this isn’t it. You can grab any other Elements book on the shelf, because they all do that.
    [Show full text]